In [2]:
!curl https://storage.yandexcloud.net/academy.ai/practica/parkinsons.data > parkinsons.data
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
100 40697 100 40697 0 0 156k 0 --:--:-- --:--:-- --:--:-- 157k
In [45]:
import pandas as pd
from xgboost import XGBClassifier
from sklearn.model_selection import train_test_split
import plotly.express as px
import plotly.graph_objs as go
from plotly.offline import download_plotlyjs, init_notebook_mode, plot, iplot
init_notebook_mode(connected=True)
import numpy
In [46]:
from ydata_profiling import ProfileReport
In [47]:
df = pd.read_csv("parkinsons.data")
In [48]:
profile = ProfileReport(df,
title='Pandas Profiling Report')
In [5]:
profile.to_widgets()
Summarize dataset: 0%| | 0/5 [00:00<?, ?it/s]
Generate report structure: 0%| | 0/1 [00:00<?, ?it/s]
Render widgets: 0%| | 0/1 [00:00<?, ?it/s]
VBox(children=(Tab(children=(Tab(children=(GridBox(children=(VBox(children=(GridspecLayout(children=(HTML(valu…
In [6]:
profile
Render HTML: 0%| | 0/1 [00:00<?, ?it/s]
Out[6]:
In [53]:
df[[ 'MDVP:Fo(Hz)', 'MDVP:Fhi(Hz)', 'MDVP:Flo(Hz)', 'MDVP:Jitter(%)',
'MDVP:Jitter(Abs)', 'MDVP:RAP', 'MDVP:PPQ', 'Jitter:DDP',
'MDVP:Shimmer', 'MDVP:Shimmer(dB)', 'Shimmer:APQ3', 'Shimmer:APQ5',
'MDVP:APQ', 'Shimmer:DDA', 'NHR', 'HNR', 'status', 'RPDE', 'DFA',
'spread1', 'spread2', 'D2', 'PPE']].corr()[['status']]
Out[53]:
| status | |
|---|---|
| MDVP:Fo(Hz) | -0.383535 |
| MDVP:Fhi(Hz) | -0.166136 |
| MDVP:Flo(Hz) | -0.380200 |
| MDVP:Jitter(%) | 0.278220 |
| MDVP:Jitter(Abs) | 0.338653 |
| MDVP:RAP | 0.266668 |
| MDVP:PPQ | 0.288698 |
| Jitter:DDP | 0.266646 |
| MDVP:Shimmer | 0.367430 |
| MDVP:Shimmer(dB) | 0.350697 |
| Shimmer:APQ3 | 0.347617 |
| Shimmer:APQ5 | 0.351148 |
| MDVP:APQ | 0.364316 |
| Shimmer:DDA | 0.347608 |
| NHR | 0.189429 |
| HNR | -0.361515 |
| status | 1.000000 |
| RPDE | 0.308567 |
| DFA | 0.231739 |
| spread1 | 0.564838 |
| spread2 | 0.454842 |
| D2 | 0.340232 |
| PPE | 0.531039 |
In [7]:
df['status'].value_counts()
Out[7]:
status 1 147 0 48 Name: count, dtype: int64
In [8]:
df.head(3)
Out[8]:
| name | MDVP:Fo(Hz) | MDVP:Fhi(Hz) | MDVP:Flo(Hz) | MDVP:Jitter(%) | MDVP:Jitter(Abs) | MDVP:RAP | MDVP:PPQ | Jitter:DDP | MDVP:Shimmer | ... | Shimmer:DDA | NHR | HNR | status | RPDE | DFA | spread1 | spread2 | D2 | PPE | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | phon_R01_S01_1 | 119.992 | 157.302 | 74.997 | 0.00784 | 0.00007 | 0.00370 | 0.00554 | 0.01109 | 0.04374 | ... | 0.06545 | 0.02211 | 21.033 | 1 | 0.414783 | 0.815285 | -4.813031 | 0.266482 | 2.301442 | 0.284654 |
| 1 | phon_R01_S01_2 | 122.400 | 148.650 | 113.819 | 0.00968 | 0.00008 | 0.00465 | 0.00696 | 0.01394 | 0.06134 | ... | 0.09403 | 0.01929 | 19.085 | 1 | 0.458359 | 0.819521 | -4.075192 | 0.335590 | 2.486855 | 0.368674 |
| 2 | phon_R01_S01_3 | 116.682 | 131.111 | 111.555 | 0.01050 | 0.00009 | 0.00544 | 0.00781 | 0.01633 | 0.05233 | ... | 0.08270 | 0.01309 | 20.651 | 1 | 0.429895 | 0.825288 | -4.443179 | 0.311173 | 2.342259 | 0.332634 |
3 rows × 24 columns
In [9]:
X = df.drop(["status","name"] ,axis=1)
y = df['status']
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.33, random_state=42, stratify=y)
In [10]:
train_vc = y_train.value_counts()
print(f"Отношение классов на тренировочной датасете {train_vc[0]/train_vc[1]}")
Отношение классов на тренировочной датасете 0.32653061224489793
In [11]:
test_vc = y_test.value_counts()
print(f"Отношение классов на тестовом датасете {test_vc[0]/test_vc[1]}")
Отношение классов на тестовом датасете 0.32653061224489793
In [12]:
xgbc = XGBClassifier()
In [13]:
xgbc.fit(X_train, y_train)
Out[13]:
XGBClassifier(base_score=None, booster=None, callbacks=None,
colsample_bylevel=None, colsample_bynode=None,
colsample_bytree=None, device=None, early_stopping_rounds=None,
enable_categorical=False, eval_metric=None, feature_types=None,
gamma=None, grow_policy=None, importance_type=None,
interaction_constraints=None, learning_rate=None, max_bin=None,
max_cat_threshold=None, max_cat_to_onehot=None,
max_delta_step=None, max_depth=None, max_leaves=None,
min_child_weight=None, missing=nan, monotone_constraints=None,
multi_strategy=None, n_estimators=None, n_jobs=None,
num_parallel_tree=None, random_state=None, ...)In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.
XGBClassifier(base_score=None, booster=None, callbacks=None,
colsample_bylevel=None, colsample_bynode=None,
colsample_bytree=None, device=None, early_stopping_rounds=None,
enable_categorical=False, eval_metric=None, feature_types=None,
gamma=None, grow_policy=None, importance_type=None,
interaction_constraints=None, learning_rate=None, max_bin=None,
max_cat_threshold=None, max_cat_to_onehot=None,
max_delta_step=None, max_depth=None, max_leaves=None,
min_child_weight=None, missing=nan, monotone_constraints=None,
multi_strategy=None, n_estimators=None, n_jobs=None,
num_parallel_tree=None, random_state=None, ...)In [14]:
xgbc.score(X_test, y_test)
Out[14]:
0.9076923076923077
In [15]:
from xgboost import train, DMatrix
trainDmatrix = DMatrix(X_train, label=y_train) #, weight=traintest_sample_weight)
validDmatrix = DMatrix(X_test, label=y_test) #, weight=valid_sample_weight)
booster = train({'eval_metric': 'auc'}, trainDmatrix, num_boost_round=100,
evals=[(trainDmatrix,'train'), (validDmatrix, 'valid')], early_stopping_rounds=50,
verbose_eval=10)
[0] train-auc:1.00000 valid-auc:0.87500 [10] train-auc:1.00000 valid-auc:0.90625 [20] train-auc:1.00000 valid-auc:0.91837 [30] train-auc:1.00000 valid-auc:0.92219 [40] train-auc:1.00000 valid-auc:0.92219 [50] train-auc:1.00000 valid-auc:0.92219 [60] train-auc:1.00000 valid-auc:0.92219 [70] train-auc:1.00000 valid-auc:0.92219 [72] train-auc:1.00000 valid-auc:0.92219
In [16]:
validDmatrix = DMatrix(X_test)
predicted = booster.predict(validDmatrix)
In [17]:
demo_test_df = X_test.copy()
demo_test_df['pred'] = predicted
demo_test_df['pred'] = demo_test_df['pred'].apply(lambda x: 1 if x >= 0.5 else 0)
demo_test_df['actual'] = y_test
Оценщик не правильно предсказал 9 примеров на тестовом наборе данных
In [18]:
demo_test_df[demo_test_df['pred'] != demo_test_df['actual']]['actual'].value_counts()
Out[18]:
actual 1 7 0 2 Name: count, dtype: int64
In [19]:
wrong_pred_df = demo_test_df[demo_test_df['pred'] != demo_test_df['actual']]
wrong_pred_df
Out[19]:
| MDVP:Fo(Hz) | MDVP:Fhi(Hz) | MDVP:Flo(Hz) | MDVP:Jitter(%) | MDVP:Jitter(Abs) | MDVP:RAP | MDVP:PPQ | Jitter:DDP | MDVP:Shimmer | MDVP:Shimmer(dB) | ... | NHR | HNR | RPDE | DFA | spread1 | spread2 | D2 | PPE | pred | actual | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 7 | 107.332 | 113.840 | 104.315 | 0.00290 | 0.00003 | 0.00144 | 0.00182 | 0.00431 | 0.01567 | 0.134 | ... | 0.00344 | 26.892 | 0.637420 | 0.763262 | -6.167603 | 0.183721 | 2.064693 | 0.163755 | 0 | 1 |
| 107 | 151.884 | 157.765 | 133.751 | 0.00258 | 0.00002 | 0.00115 | 0.00122 | 0.00346 | 0.01450 | 0.137 | ... | 0.00659 | 26.833 | 0.393563 | 0.700246 | -6.600023 | 0.249172 | 2.499148 | 0.105306 | 0 | 1 |
| 183 | 117.226 | 123.925 | 106.656 | 0.00417 | 0.00004 | 0.00186 | 0.00270 | 0.00558 | 0.01909 | 0.171 | ... | 0.00955 | 23.079 | 0.603515 | 0.669565 | -5.619070 | 0.191576 | 2.027228 | 0.215724 | 1 | 0 |
| 177 | 151.737 | 190.204 | 129.859 | 0.00314 | 0.00002 | 0.00135 | 0.00162 | 0.00406 | 0.01469 | 0.132 | ... | 0.00570 | 24.151 | 0.396610 | 0.745957 | -6.486822 | 0.197919 | 2.449763 | 0.132703 | 0 | 1 |
| 105 | 149.689 | 160.368 | 133.608 | 0.00257 | 0.00002 | 0.00116 | 0.00134 | 0.00349 | 0.01346 | 0.126 | ... | 0.00578 | 27.421 | 0.365488 | 0.719974 | -6.997403 | 0.210279 | 1.994387 | 0.112878 | 0 | 1 |
| 188 | 114.563 | 119.167 | 86.647 | 0.00327 | 0.00003 | 0.00146 | 0.00184 | 0.00439 | 0.01185 | 0.106 | ... | 0.00703 | 24.775 | 0.555303 | 0.659132 | -6.710219 | 0.149694 | 1.913990 | 0.121777 | 1 | 0 |
| 113 | 210.141 | 232.706 | 185.258 | 0.00534 | 0.00003 | 0.00321 | 0.00280 | 0.00964 | 0.01680 | 0.149 | ... | 0.00620 | 23.671 | 0.441097 | 0.722254 | -5.963040 | 0.250283 | 2.489191 | 0.177807 | 0 | 1 |
| 28 | 155.358 | 227.383 | 80.055 | 0.00310 | 0.00002 | 0.00159 | 0.00176 | 0.00476 | 0.01718 | 0.161 | ... | 0.00677 | 25.970 | 0.470478 | 0.676258 | -7.120925 | 0.279789 | 2.241742 | 0.108514 | 0 | 1 |
| 118 | 178.285 | 442.824 | 82.063 | 0.00462 | 0.00003 | 0.00157 | 0.00194 | 0.00472 | 0.01279 | 0.129 | ... | 0.00856 | 25.020 | 0.470422 | 0.655239 | -4.913137 | 0.393056 | 2.816781 | 0.251972 | 0 | 1 |
9 rows × 24 columns
In [24]:
import shap
%matplotlib inline
In [25]:
explainer = shap.TreeExplainer(booster)
shap_values = explainer.shap_values(X_test)
In [26]:
shap.summary_plot(shap_values, X_test, plot_type="bar")
In [27]:
shap.summary_plot(shap_values, X_test)
In [28]:
# same shap.plots.scatter(shap_values[:, "PPE"], color=shap_values)
shap.dependence_plot("PPE", shap_values, X_test)
In [29]:
shap_values = explainer(X_test)
shap.plots.waterfall(shap_values[0])
In [30]:
shap.plots.force(shap_values[0])
Out[30]:
Visualization omitted, Javascript library not loaded!
Have you run `initjs()` in this notebook? If this notebook was from another user you must also trust this notebook (File -> Trust notebook). If you are viewing this notebook on github the Javascript has been stripped for security. If you are using JupyterLab this error is because a JupyterLab extension has not yet been written.
Have you run `initjs()` in this notebook? If this notebook was from another user you must also trust this notebook (File -> Trust notebook). If you are viewing this notebook on github the Javascript has been stripped for security. If you are using JupyterLab this error is because a JupyterLab extension has not yet been written.
In [31]:
shap.plots.force(shap_values[:500])
Out[31]:
Visualization omitted, Javascript library not loaded!
Have you run `initjs()` in this notebook? If this notebook was from another user you must also trust this notebook (File -> Trust notebook). If you are viewing this notebook on github the Javascript has been stripped for security. If you are using JupyterLab this error is because a JupyterLab extension has not yet been written.
Have you run `initjs()` in this notebook? If this notebook was from another user you must also trust this notebook (File -> Trust notebook). If you are viewing this notebook on github the Javascript has been stripped for security. If you are using JupyterLab this error is because a JupyterLab extension has not yet been written.
In [32]:
shap.plots.beeswarm(shap_values)
In [33]:
Проверка/визуализация примеров которые были определены с ошибкой.
In [38]:
from sklearn.manifold import TSNE
In [39]:
#data = df[~df.index.isin(wrong_pred_df.index)].copy().reset_index(drop=True)
data = df.copy()
data.loc[data.index.isin(wrong_pred_df.index), 'status'] = data.loc[data.index.isin(wrong_pred_df.index), 'status'] + 2
In [40]:
data.drop(['name'], axis=1, inplace=True)
y_data = data[["status"]]
data.drop(['status'], axis=1, inplace=True)
In [41]:
tsne = TSNE(
perplexity=30,
metric="euclidean",
n_jobs=8,
random_state=42,
verbose=True,
)
In [42]:
%time
embedding_train = tsne.fit_transform(data)
CPU times: user 2 µs, sys: 1e+03 ns, total: 3 µs Wall time: 7.87 µs [t-SNE] Computing 91 nearest neighbors... [t-SNE] Indexed 195 samples in 0.001s... [t-SNE] Computed neighbors for 195 samples in 0.011s... [t-SNE] Computed conditional probabilities for sample 195 / 195 [t-SNE] Mean sigma: 16.717366 [t-SNE] KL divergence after 250 iterations with early exaggeration: 46.679577 [t-SNE] KL divergence after 1000 iterations: 0.141709
In [43]:
fig = px.scatter(
embedding_train, x=0, y=1,
color=y_data['status'],
color_discrete_sequence=["red", "green", "blue", "goldenrod"]
)
fig.show()
# 0 = 0
# 1 = 1
# 2 = 0 - должен быть 1
# 3 = 1 - должен быть 0
In [ ]:
In [ ]: